List your company

Free-format Multimodal Hallucination Assessment on RefoMB (dev)

62.9TrustOmniLMM + RLAIF-V BoN
24.83634.71844.654.482May 27, 2024

Evaluation results

MethodLinks
OmniLMM + RLAIF-V BoNSize=12B, Feedback=self2024.0562.930.3
OmniLMM + RLAIF-VSize=12B, Feedback=self2024.0558.128.3
LLaVA 1.5 + RLAIF-V BoNSize=7B, Feedback=LLaVA-NeXT2024.0555.724.4
Less-is-moreSize=7B, Feedback=X2024.0551.116.2
MiniGeminiSize=34B, Feedback=X2024.055036.9
GPT-4VSize=Unknown, Feedback=Unknown2024.055050
LLaVA 1.5 + RLAIF-VSize=7B, Feedback=LLaVA-NeXT2024.0547.222.5
OmniLMMSize=12B, Feedback=X2024.0544.718.5
LLaVA-NeXTSize=34B, Feedback=X2024.0544.435.4
POVIDSize=7B, Feedback=Rule2024.0544.413.6
CCA-LLaVASize=7B, Feedback=X2024.0541.921.7
RLHF-VSize=13B, Feedback=Human2024.0541.417.7
Qwen-VL-ChatSize=10B, Feedback=X2024.0540.917.7
VCDSize=7B, Feedback=X2024.0539.916.7
HA-DPOSize=7B, Feedback=Rule2024.0539.917.2
SilkieSize=10B, Feedback=GPT-4V2024.0538.921.2
LLaVA 1.5Size=7B, Feedback=X2024.0536.916.2
OPERASize=7B, Feedback=X2024.0533.813.1
AMP-MEGSize=13B, Feedback=Rule2024.0530.314.6
LLaVA-RLHFSize=13B, Feedback=Human2024.0526.317.2